This invention pertains to novel methods for high-throughput screening for pharmaceutical compounds, in particular those that bind to RNA sequences involved in the pathogenesis of disease or in regulation of a physiological function.
Pharmaceuticals can be developed from lead compounds that are identified through a random screening process directed towards a target, such as a nucleic acid or a protein receptor. Large scale screening approaches can be complicated by a number of factors. First, many assays are laborious or expensive to perform. Assays may involve experimental animals, cell lines, or tissue cultures that are difficult or expensive to acquire or maintain. These considerations often place practical limitations on the number of compounds that reasonably can be screened. Thus, those employing random screening methods are frequently forced to limit their search to those compounds for which some prior knowledge suggests that the compounds are likely to be effective. This strategy limits the range of compounds tested, and many useful drugs may be overlooked.
Furthermore, the specificity of many biochemical assays may exclude a wide variety of useful chemical compounds, because the interactions between the ligand and the target are outside the scope of the assay. With such a specific assay, many potential pharmaceuticals may not be detected.
Finally, in most existing biochemical screening approaches to drug discovery, the system in question must be well-characterized before screening can begin. Consequently, biochemical screening for therapeutic drugs directed against many targets must await detailed biochemical characterization, a process that generally requires extensive research.
The present invention pertains specifically to the use of RNA targets in high-throughput screening methods for identification of useful ligands. The invention takes advantage of the existence of higher-order structures in naturally-occurring and synthetic RNA molecules. For example, RNA exists in both single stranded and helical duplex forms. These duplexes may be distorted by loops, bulges, base triples and junctions between helices. The structural diversity of RNA is far greater than that of DNA, and similar to that of proteins, making RNA a likely target for binding of small molecules (reviewed in Wyatt and Tinoco. The RNA World, Gesteland and Atkins, eds., Cold Spring Harbor, 1993, pp. 465-496-A.
Small molecules can bind RNA with high affinity and specificity and can block essential functions of the bound RNA. The best example of such molecules are antibiotics such as erythromycin and aminoglycosides. The first suggestion that some antibiotic translation inhibitors interact specifically with RNA was the genetic mapping of resistance to kanamycin and gentamicin to the methylation of 16S RNA (Thompson et al., Mol. Gen. Genet. 201:168, 1985). Erythromycin binds to bacterial RNA and releases peptidyl-tRNA and mRNA (Menninger et al., Mol. Gen. Genet. 243:225, 1994). 2-DOS-containing aminoglycosides bind specifically to the structures of HIV RNA known as the RRE, block binding of the HIV Rev protein to this RNA, and thereby inhibit HIV replication in tissue culture cells (Zapp et al., Cell 74:969, 1993). In addition, although aminoglycosides have long been developed as translation inhibitors, they were only recently shown to bind to rRNA in the absence of proteins (Purohit and Stern, Nature 370:659, 1994). Hygromycin B inhibits coronaviral RNA synthesis and is thought to do so by binding to the viral RNA and blocking specifically the translation of viral RNA (Macintyre et al., Antimicrob. Agents Chemother. 35:2630, 1991).
Existing assays for ligands of nucleic acids, such as, for example, methods that use equilibrium dialysis, differential scanning calorimetry, viscometric analyses, or UV melting, have not been used in high-throughput applications. Thus, prior to the present invention, random screening approaches for non-oligonucleotide ligands of RNA were limited to compounds for which some prior knowledge suggested that they might be effective. This strategy has been successful (Zapp et al., 1993), but is limited in the range of compounds that can be tested on a practical scale.
U.S. Pat. No. 5,270,163 describes the SELEX system for the identification of oligonucleotides that bind specific targets. In this system, random oligonucleotides are affinity-selected and amplified, followed by several cycles of re-selection and amplification. This method, however, is limited to screening for oligonucleotide ligands and cannot be applied in reverse, i.e., to search for non-oligonucleotide ligands that bind to nucleic acids.
U.S. Pat. No. 5,306,619 discloses a screening method to identify compounds that bind particular DNA target sequences. In this method, a test nucleic acid is constructed in which the target sequence is placed adjacent to a known protein-binding DNA sequence. The effect of test compounds on the binding of the cognate protein to the protein-binding DNA sequence is then measured. This method requires conditions in which melting of DNA hybrids and unfolding of DNA structure do not occur. RNA, by contrast, can undergo much more dramatic variations in patterns of base-pairing and overall conformation.
Thus, there is a need in the art for efficient and cost-effective high-throughput methods for random screening of large numbers of non-oligonucleotide small molecules for their ability to bind physiologically, medically, or commercially significant RNA molecules.
The present invention encompasses high-throughput screening methods to identify ligands that bind any predetermined target RNA. The methods are carried out by the steps of: selecting as test ligands a plurality of compounds not known to bind to the target RNA sequence; incubating the target RNA sequence in the presence of each of the test ligands to produce a test combination; incubating the target RNA sequence in the absence of a test ligand to produce a control combination; measuring the conformation of the target RNA in each combination; selecting as a ligand any test ligand that causes a measurable change in the target RNA conformation in the test combination relative to the control combination; and repeating the method with a plurality of said test ligands to identify a ligand that binds to the target RNA sequence. Ligands identified by the methods of the present invention may cause the target RNA to change from a less folded to more folded conformation, from a more folded to less folded conformation, or from a first folded conformation to a second, alternative, folded conformation. Furthermore, the test and control combinations may be subjected to conditions that, in the absence of ligands (i.e., in the control combination), denature a detectable fraction of the target RNA.
In practicing the present invention, the effect of test ligands on the folding state of the target RNA is determined using well-known methods, including without limitation hybridization with complementary oligonucleotides, treatment with conformation-specific nucleases, binding to matrices specific for single-stranded or double-stranded nucleic acids, and fluorescence energy transfer between fluorescence probes. In one embodiment, the target RNA is radiolabelled and incubated with a biotinylated oligonucleotide that preferentially hybridizes to a particular conformation of the target RNA; following capture of biotinylated molecules using immobilized streptavidin, the extent of hybridization can be readily quantified by measurement of immobilized radiolabel. In another embodiment, the target RNA contains two different fluorescence probes in which the fluorescence emission wavelength maximum of the first probe overlaps the fluorescence absorption maximum of the second probe. The probes are positioned within the target RNA so that the efficiency of fluorescence energy transfer between the probes is dependent upon the target RNA conformation.
A xe2x80x9cmeasurable changexe2x80x9d in target RNA conformation as detected by any of the above or other methods is one in which the difference in the measured parameter between the test and control combinations is greater than that expected due to random statistical variation.